Hazardous reactive chemicals are commonly used as reactants to promote chemical vapor deposition (CVD), atomic layer deposition (ALD), and other processes. Some of these chemicals are particularly unstable and dangerous. For example, CVD and ALD growth of a variety of nitride films have been demonstrated with the aid of hydrazine (N2H4) as the nitrogen and reducing source with superior process performance and film properties over other, less reactive nitrogen sources. Nevertheless, the use of hydrazine in commercial CVD and ALD processes has been hampered by concerns about the safety implications of this extremely unstable, explosive, and hazardous chemical reactant.
Generally speaking, hydrazine is a low cost chemical widely used in the chemical industry for a variety of processes. Hydrazine and hydrazine derivatives are also being used as rocket propellants. Overall annual production of hydrazine typically exceeds 50,000 tons. With the exception of propellants, hydrazine is typically stored, transported, and used in the much safer aqueous solutions ranging from 35-64 weight-percent (wt %). For applications such as CVD and ALD, however, these aqueous solutions are typically disadvantageous. In particular, most applications require high-purity anhydrous hydrazine with its associated safety and toxicity issues. A spill of liquid anhydrous hydrazine may easily self ignite, raise the localized temperature and, as a result, the localized partial pressure, and ultimately detonate when the localized partial pressure in air exceeds 4.7% and the localized temperature exceeds 92 degrees Celsius (° C.). The explosion damage coupled with the spread of toxic NO and NO2 by-products can be catastrophic. The potential danger from anhydrous hydrazine spills makes it unpopular as a rocket propellant, let alone as an industrial chemical reactant. Accordingly, hydrazine derivatives such as monomethylhydrazine and dimethylhydrazine have been invoked as safer substitutions. Unfortunately, both derivatives have found very limited use in processes such as CVD and ALD due to their lower reactivity and the tendency to include carbon in films.
Hydrazine cyanurate (HC) is a stable complex of hydrazine that was developed and made public by Nachbaur and Leiseder for the purpose of purification and separation as early as 1971. When heated above 170° C., HC decomposes into the ingredients hydrazine and cyanuric acid (CA). Hydrazine production from HC depends on the temperature and the quantity of HC. In the temperature range between 170-245° C., the partial pressure of CA is negligible. Above 245° C., CA disadvantageously decomposes into hydrocyanic acid. Otherwise, when HC is relatively cold (i.e. below about 100° C.), it is stable and safe. These characteristics led others to propose HC for industrial hydrazine production purposes. More particularly, T. J. Kropewnicki et al. in “Hydrazine Cyanurate as a Nitrogen Source for Thin Nitride Film Growth,” Journal of Vacuum Science and Technology A16(1), January/February 1998, pp. 139-144; and K. J. Taylor et al. in U.S. Pat. No. 6,465,350, entitled “Aluminum Nitride Film Formation on Integrated Circuits,” propose forming hydrazine from HC for CVD nitridization reactions. Nevertheless, these references only set forth methodologies consisting of heating HC to an elevated temperature (e.g., 210° C.) and simply leaving it at this temperature while one or several CVD reaction sequences are allowed to run. No regard is paid to the consistency of the hydrazine supply within a given CVD sequence or from one sequence to the next, to the potential for the catastrophic accumulation of hydrazine in the reactor or the heated source due to overproduction and condensation, or to compensating for the drop in hydrazine production as hydrazine in the HC source is exhausted. As a result, even with these various attempts to harness hydrazine from HC, HC has seen no significant implementation in industrial deposition processes despite the many well known benefits of using hydrazine as a reactant and despite the availability of HC for over 35 years.
Low-volatility chemicals can be consistently maintained at a desired partial pressure when placed in an elevated temperature container. This stable vapor pressure at a given temperature ensures the consistency of chemical supply and an automatic, per-demand, supplementation of the source to maintain the desired pressure. For example, during processing, the source chemical is being drawn into the process chamber to effectively drive the partial pressure down and subsequently drive more chemical into the gas phase to maintain the equilibrium vapor pressure.
Evolving hydrazine from HC is fundamentally different. Hydrazine evolution from HC is a continuous process that produces substantial quantities of hydrazine for as long as the source temperature is sufficiently high. For example, about 1.8×10−8 grams/sec of hydrazine is produced per each gram of HC at 170° C., as detailed below. FIG. 1 displays hydrazine production versus time, normalized to one gram of HC, as measured from the pressure rise inside a ˜2 liter (L) container. Clearly, the rate of hydrazine production increases at higher temperatures. The absolute rates of hydrazine production are estimated from the given known volume and the potency of HC. These rates were used to extract the kinetics of hydrazine production, as displayed in FIG. 2. A good fit to a linear dependence of log(rate) versus 1/T, wherein T is the absolute temperature, indicates a first order Arrhenius behavior with an activation energy of 48±2 kilocalories per mole and the normalized rate is given by: k (hydrazine production/gram of HC)=6±2×1015 exp(−(24,000±1,000)/T) grams/sec. Accordingly, significant production of hydrazine does not start at temperatures lower than about 170° C. At room temperature (i.e., 25° C.), for example, hydrazine production is negligibly low at 4×10−23 grams/sec/gram-of-HC. At that rate, one kilogram (kg) of HC will require almost 800 years to produce a negligible one nanogram of hydrazine. However, at 200° C., one kg produces 550 micrograms in one second. At 230° C., which is recommended as the highest safe temperature for HC heating (with ˜15° C. margin), hydrazine production from one kg of HC reaches ˜40 grams/hour.
Production of hydrazine at any given temperature is clearly a continuous process that declines with the exhaustion of the HC load. Since hydrazine is being continuously produced, it clearly lacks any mechanism to ensure that this hazardous chemical is not over-produced and consequently accumulates beyond demand. Eventually, over-production of hydrazine may exceed the dew point at the given source temperature and excess hydrazine may condense into liquid anhydrous hydrazine. For example, if the temperature of a 1 kg HC load is maintained at 200° C., the initial production rate of ˜0.0330 grams/min will exceed the dew-point pressure of 55 Torr after approximately 15 minutes within a 5 L source chamber maintained at 50° C. At that point the source will start accumulating hazardous liquid anhydrous hydrazine. Indeed, the inventors observed that hydrazine production did not produce any pressure rise above ˜18 Torr when the source chamber temperature was maintained at 32° C., suggesting condensation. Upon heating the source chamber to 40° C., the pressure instantaneously rose to ˜28 Torr and then saturated there. Although production rate will slow down with the exhaustion of HC, it will produce ten grams of liquid anhydrous hydrazine within approximately five hours. This over-production of hydrazine may defeat the purpose of using HC as a means for on-demand production of hydrazine. Likewise, there is no mechanism to ensure adequate supplementation of hydrazine to compensate for process usage. Hydrazine production at any given temperature scales with the potency of HC which inevitably declines as the HC adduct (i.e., chemical complex) breaks down into hydrazine and CA. Therefore, if the HC temperature is minimized to minimize over production during idle time, it may soon be too low for consistent hydrazine supply during processing, especially since the potency of HC will decline during usage as explained below.
To further demonstrate the major drawbacks of prior art practice wherein HC is maintained at a constant temperature, the inventors monitored the pressure of hydrazine inside a 2 L source chamber during a TiN ALD process using a “Synchronous Modulation of Flow and Draw” ALD process in accordance with aspects of U.S. Pat. No. 6,911,092, entitled “ALD Apparatus and Method,” to one of the presently named inventors, which is hereby incorporated by reference herein. Hydrazine was produced from HC at a controlled temperature of 185° C. The TiN process applied saturating TiCl4 and hydrazine pulses to grow a high quality film at a 270° C. process temperature. Likewise, a 50 milliseconds (msec) pulse of hydrazine from a source pressure of 8 Torr generated a dose of ˜50 μgrams of hydrazine, which was sufficient to saturate the hydrazine reaction. FIG. 3 (solid line) displays the pressure of hydrazine within the source chamber during a first idle time, a first process wherein the hydrazine was pulsed for 50 msec, a second idle time, a second process wherein the hydrazine was pulsed for 100 msec, a third idle time, and a third process wherein the hydrazine was pulsed for 200 msec. In each process, the TiCl4 was pulsed for 30 msec, which was sufficient to achieve saturation. FIG. 3 demonstrates that the practice of the prior art is neither safe nor suitable for consistent delivery of hydrazine. Hydrazine production was continuous during idle-times. During the process periods, the production of hydrazine clearly lagged behind the utilization as indicated by the dramatic pressure drops. In fact, the dose/cycle varied between 105 and 5 μgrams/pulse, 385 and 33 μgrams/pulse, and 736 and 7 μgrams/pulse over the first, second and third process runs, respectively, making any attempt to use the prior art practice for processing hopeless. There was also evidence for declining production rates with the usage of HC. For example, the average production rate declined from 1.0275×10−7 grams/sec/gram-of-HC during the first idle time to 1.0230×10−7 and 1.0216×10−7 during the second and third idle times, respectively. That was a 0.4% and 0.6% rate decrease for relatively short processes running consecutively.
In summary, a hydrazine supply, when simply held at a constant temperature, may vary by a great factor given the dependence on multiple parameters such as HC potency, initial load, temperature, the extent of previous idle time where hydrazine was produced but not consumed, and the extent of previous processing wherein hydrazine may have been consumed faster than produced. These variations could easily span an order of magnitude as shown in the example of FIG. 3. At that level of source inconsistency, even state of the art chemical delivery instruments such as mass flow controllers (MFC) or pressure controllers (PRC) might be driven into instability. Given the lack of consistency, and the lack of dependable mechanisms to prevent over production and to react to process related withdrawal or idle-time related consumption halts, the simple constant-temperature HC source of the prior art is neither safe nor suitable for process applications such as CVD and ALD.
Therefore, there is a need for apparatus and methods that can harness relatively stable, easily-handled chemical complexes of hazardous reactants such as HC into safe and consistent on-demand chemical sources for processes such as CVD, ALD, surface preparation, surface and bulk treatment, and reactive abatement.